(1) Field of the Invention
The present invention relates to light collection efficiency and package yield improvements for the optical structure and microelectronic fabrication process of semiconductor color imaging devices.
(2) Description of Prior Art
Synthetic reconstruction of color images in solid-state analog or digital video cameras is conventionally performed through a combination of an array of optical microlens and spectral filter structures and integrated circuit amplifier automatic gain control operations following a prescribed sequence of calibrations in an algorithm.
Typically solid-state color cameras are comprised of charge-coupled device (CCD), Charge-Injection Device (CID), or Complementary Metal-Oxide Semiconductor (CMOS) structures with planar arrays of microlenses and primary color filters mutually aligned to an area array of photodiodes patterned onto a semiconductor substrate. The principal challenge in the design of solid-state color camera devices is the trade-off between adding complexity and steps to the microelectronic fabrication process wherein color filters are integrally formed in the semiconductor cross-sectional structure versus adding complexity and integrated electronic circuitry for conversion of the optical analog signals into digital form and signal processing with color-specific automated gain-control amplifiers requiring gain-ratio balance. The trade-off between microelectronic fabrication process complexity versus electronic complexity is determined by a plurality of factors, including product manufacturing cost and optoelectronic performance.
Color-photosensitive integrated circuits require carefully configured color filters to be deposited on the upper layers of a semiconductor device in order to accurately translate a visual image into its color components. Conventional configurations may generate a color pixel by employing four adjacent pixels on an image sensor. Each of the four pixels is covered by a different color filter selected from the group of red, blue and two green pixels, thereby exposing each monochromatic pixel to only one of the three basic colors. Simple algorithms are subsequently applied to merge the inputs from the three monochromatic pixels to form one full color pixel. The color filter deposition process and its relationship to the microlens array formation process determine the production cycle-time, test-time, yield, and ultimate manufacturing cost. It is an object of the present invention to teach color-filter processes which optimize these stated factors without the microlens array(s) and the associated complex process steps.
While color image formation may be accomplished by recording appropriately filtered images using three separate arrays, such systems tend to be large and costly. Cameras in which a full color image is generated by a single detector array offer significant improvements in size and cost but have inferior spatial resolution. Single-chip color arrays typically use color filters that are aligned with individual columns of photodetector elements to generate a color video signal. In a typical stripe configuration, green filters are used on every other column with the intermediate columns alternatively selected for red or blue recording. To generate a color video signal using an array of this type, intensity information from the green columns is interpolated to produce green data at the red and blue locations. This information is then used to calculate a red-minus-green signal from red-filtered columns and a blue-minus-green signal from the blue ones.
Complete red-minus-green and blue-minus-green images are subsequently interpolated from this data yielding three complete images. Commercial camcorders use a process similar to this to generate a color image but typically utilize more complicated mosaic-filter designs. The use of alternate columns to yield color information decreases the spatial resolution in the final image.
The elementary unit-cell of the imager is defined as a pixel, characterized as an addressable area element with intensity and chroma attributes related to the spectral signal contrast derived from the photon collection efficiency. Prior art conventionally introduces a microlens on top of each pixel to focus light rays onto the photosensitive zone of the pixel.
The optical performance of semiconductor imaging arrays depends on pixel size and the geometrical optical design of the camera lens, microlenses, color filter combinations, spacers, and photodiode active area size and shape. The function of the microlens is to efficiently collect incident light falling within the acceptance cone and refract this light in an image formation process onto a focal plane at a depth defined by the planar array of photodiode elements. Significant depth of focus may be required to achieve high resolution images and superior spectral signal contrast since the typical configuration positions the microlens array at the top light collecting surface and the photosensors at the semiconductor substrate surface.
When a microlens element forms an image of an object passed by a video camera lens, the amount of radiant energy (light) collected is directly proportional to the area of the clear aperture, or entrance pupil, of the microlens. At the image falling on the photodiode active area, the illumination (energy per unit area) is inversely proportional to the image area over which the object light is spread. The aperture area is proportional to the square of the pupil diameter and the image area is proportional to the square of the image distance, or focal length. The ratio of the focal length to the clear aperture of the microlens is known in Optics as the relative aperture or f-number.
The illumination in the image arriving at the plane of the photodetectors is inversely proportional to the square of the ratio of the focal length to clear aperture. An alternative description uses the definition that the numerical aperture (NA) of the lens is the reciprocal of twice the f-number. The concept of depth of focus is that there exists an acceptable range of blur (due to defocussing) that will not adversely affect the performance of the optical system. The depth of focus is dependent on the wavelength of light, and, falls off inversely with the square of the numerical aperture. Truncation of illuminance patterns falling outside the microlens aperture results in diffractive spreading and clipping or vignetting, producing undesirable nonuniformities and a dark ring around the image.
The limiting numerical aperture or f-stop of the imaging camera's optical system is determined by the smallest aperture element in the convolution train. Typically, the microlens will be the limiting aperture in video camera systems. Prior Art is characterized by methods and structures to maximize the microlens aperture by increasing the radius of curvature, employing lens materials with increased refractive index, or, using compound lens arrangements to extend the focal plane deeper to match the multilayer span required to image light onto the buried photodiodes at the base surface of the semiconductor substrate. Light falling between photodiode elements or on insensitive outer zones of the photodiodes, known as dead zones, may cause image smear or noise. With Industry trends to increased miniaturization, smaller photodiodes are associated with decreasing manufacturing cost, and, similarly, mitigate against the extra steps of forming layers for Prior Art compound lens arrangements to gain increased focal length imaging. Since the microlens is aligned and matched in physical size to shrinking pixel sizes, larger microlens sizes are not a practical direction. Higher refractive index materials for the microlens would increase the reflection-loss at the air-microlens interface and result in decreased light collection efficiency and reduced spectral signal contrast or reduced signal-to-noise ratio. Limits to the numerical aperture value of the microlens are imposed by the inverse relationship of the depth of focus decreasing as the square of the numerical aperture, a strong quadratic sensitivity on the numerical aperture.
Typically, a pixel with a microlens requires a narrower incident light angle than a pixel that does not use a microlens, imposing additional optical design implications for the lens of the camera.
The design challenge for creating superior solid-state color imagers is, therefore, to optimize spectral collection efficiency to maximize the fill-factor of the photosensor array elements without vignetting (losses from overfilling) and associated photosensor cross-talk, and, with the minimum number of microelectronic fabrication process steps. The present invention is clearly distinguished from Prior Art by introducing at least one high transmittance planar film-layer of specified optical and physical properties directly over color-filters without the use of microlens arrays.
This distinction will be further demonstrated in the following sections by describing the specific related optical conditions to be satisfied at the interfaces between the functional layers comprising the semiconductor color-imaging device when no microlenses are used.
On colors only products where no microlens layer is formed, the color pixel surface is not flat. The curvature of the color filter surface will cause incident image light to refract and the image position and power-density (viz., irradiance distribution) at the sensor surface will be changed. These factors could have an effect on pixel sensitivity, signal contract and pixel cross-talk. In the colors only process, the final product wafer suffers significant topography step-height variations. During the package dicing step, residue particles remain embedded as a result of the topographical problem. The resulting entrapped residue particles impact the image quality and cause yield loss of CMOS/CCD image sensor products.
FIG. 1 exhibits the conventional Prior Art vertical semiconductor cross-sectional profile and optical configuration for color image formation. Microlens 1 residing on a planarization layer which serves as a spacer 2 collects a bundle of light rays from the image presented to the video camera and converges the light into focal cone 3 onto photodiode 8 after passing through color filter 4 residing on planarization layer 5, passivation layer 6, and metallization layer 7.
The purpose of the microlens' application in CCD and CMOS imaging devices is to increase imager sensing efficiency. FIG. 2 illustrates the geometrical optics for incident image light 9 converged by microlens element 10, color filter 11, into focal cone 12, to the focal area 13 within a photoactive area 14 surrounded by a dead or non-photosensitive area 15, wherein the sum of the areas of 14 and 15 comprise the region of the pixel.
Otsuka in U.S. Pat. No, 6,040,591 teaches a charge-coupled device (CCD) imaging array having a refractive index adjusting and planarizing layer over a microlens array layer to correct for non-normal angles of incidence affect on the image light convergence positions at the photosensor planar array and for interfacial reflection loss at the microlens surface. Otsuka assumed a typical refractive index value of n=1.75 for a reflowed polyimide resin microlens end selected a fluororesin from Asahi Glass Co., Ltd of refractive index n=1.34 for the index adjusting layer. That is, Otsuka uses an index of refraction for the refractive index adjusting layer which is lower than the microlens' index to assure bending image light rays inward toward the surface-normal to obviate vignetting at the sensor active area. FIG. 3 shows the CCD cross-sectional structure of the preferred embodiment of Otsuka's referenced patent, comprised of a photodiode 28, charge transfer portion 17, formed in a semiconductor substrate 16, having a vertical transfer electrode 18, a light shielding film 19 covering the vertical transfer electrode 18, a transparent flattening film 20 covering the photodiode 28 and light shield 19, a color filter 21 formed on the flattening film 20, a flattening film 22 formed on the color filter 21, a hemispherical microlens 23 formed on the transparent flattening film 22, and a transparent film 24 having refractive index lower than that of the microlens formed to cover the microlens. A final optional top-surface antireflection coating 25 is then formed on the film 24. Incident light, L, is shown to converge at the new, deeper focal point F, instead of the unadjusted shallower value of f0 which occurs when the index-adjusting film 24 is absent. It is noted, then, that the indices of refraction and all the prescribed layer thicknesses taught by Otsuka in the referenced patent correspond to optical designs accommodating the geometric and physical optical characteristics of the formed microlens, not those of the color filter layer(s). No special treatment or specified conditions are provided for adjustment of the planarizing spacer layer 22, nor are interface conditions between the color filter layers 21 and planarizing spacer layer 22 addressed.
The case of no microlens is not considered by Otsuka. Otsuka does consider using the index-adjusting layer as a transparent sealing resin which can be hardened and used to seal the solid-state imager as a package. It is noted that any contaminants captured in the microlens interstices will not be removed in a final cleaning process step, but will be sealed in as well. Results of embedded particulates will lead to light scattering noise effects.
An alternative approach to microlens optics and device cross-sectional adaptations, using refractive index structures configured to collect and converge image light onto the photodetecting surface of the pixel, is given by Furumiya in U.S. Pat. No. 5,844,290. It is noted that color filters, color image formation processes, and whether there is compatibility of Furiyama's structures with color filters are not discussed in Furiyama's referenced patent.
According to FIG. 4 in U.S. Pat. No. 5,844,290 by Furiyama, a preferred embodiment for the solid-state imager is comprised of a CCD structure formed of n-type silicon substrate 30, p-well 31, silicon-oxide film 38, in which are patterned n-type buried channel layer 34 above p-type layer 35, a pn junction photodiode of p+ type layer 33 above n-type layer 32 with p+ device isolation 36, and, device opening 42 and reading gate 34. Built up above the pn junction are transfer electrode 39, silicon oxide film 40, light shield film 41, insulator film 43, and, a first region of planarizing resin 45 vertically contiguous with a second region of planarizing resin layer 44, forming a top surface plane for microlens array 46.
The geometrical optics for capturing and converging image light to the photosensor plane of the CCD is depicted by normal incident light I gathered in a focal cone of the microlens. The extreme rays are refracted by the second (vertical) region of planarizing resin layer 44 into the first region of planarizing resin layer 45, to a focal point in proximity to the photodiode surface. The first region 45 is in the form of a cylindrical column and is positioned between the n-type layer 32 and a center portion of the microlens 46. The second region 44 surrounding the first region 45 has a refractive index larger than a refractive index of the first region, assuring the image light bends inwards toward the surface normal. This coaxial cylindrical arrangement can, as Furumiya states, be subject to reflection losses at the boundary between the planarizing resin layers. It is noted here for the Furumiya referenced patent, as well as we noted earlier for the Otsuko referenced patent, that the case of no microlens is not addressed.
U.S. Pat. No. 5,691,548 to Akio addresses the long focal length, film stack thickness, and vignetting problems common in Prior Art by introducing a compound lens arrangement comprised of a first positive or converging convex element in tandem with a negative or diverging (concave upward) second element. The principal problem Akio addresses is for low light levels the camera's aperture stop must be fully opened. Obliquely incident light rays will noticeably increase in their proportion to the total amount of all incident image light. Under these conditions, conventional solid-state imagers will truncate or vignette significantly, diminishing their optical sensitivity.
To solve this problem of conventional imagers not collecting and imaging light efficiently when the aperture is open fully, Akio teaches an optical arrangement so that a concave type microlens layer operates to collimate light rays collected by the convex lens so as to converge on the photosensor plane. The color image formation process and the case of no microlens is not addressed in the referenced Akio patent.
In U.S. Pat. No. 6,091,093 to Kang et al, an MOS semiconductor imager and microlens process is taught. In particular, embodiments of the invention are directed to create a number of gate islands electrically insulated from each other with spacers. The processes disclosed aims to integrate logic IC fabrication with photosensors. Conventional processes for polycide-gate or salicide-gate MOS devices generally introduce the problem of inherently forming opaque regions preventing image light from entering the photosensitive regions of the silicon at a distance below the surface. Kang et al teach a process for photocell construction without the conventional additional mask step to prevent the formation of the silicide over those silicon regions that are patterned for photodetectors. Spacers are formed above the pn-junction of the photodiode array elements such that incident light passes through the spacers and into the photosensitive region. As noted previously, Kang does not address the color formation process and his optical arrangements will not operate without a converging microlens.
The color filter process and optical film structures taught in the present invention are clearly distinguished from the Prior Art by eliminating microlenses, and, are shown to include fewer process steps with improved package final product yield.
A principal object of the present invention is to teach the method and structures for adding a specified planarization layer after the final color filter layer formation in the colors only product in which there are no microlenses. Experiments conducted by the inventors have demonstrated that the present invention improves pixel sensitivity and reduces the package yield loss through the reduction of residual trapped particulates induced in the package dicing and cleaning steps. It is an object of the present invention to reduce interfacial reflection losses and vignetting of image light by disclosing a method, structures and optical properties required for refractive index boundary-engineering.
Another object of the present invention is to provide an adaptive process wherein antireflection and image-forming structures, spectral color filters, and, combinations or varying configurations of semiconductor vertical profiles can be integrated with the result of maximizing collection efficiency of image intensity patterns on the photodiode planar arrays to achieve optimum pixel resolution and color signal contrast with minimal smear and pixel cross-talk.
In accord with a principal object of the present invention, there is provided by the present invention a manufacturing method and microelectronic fabrication process sequence which minimizes the number and task-times of the operational steps required in the production of semiconductor arrays for color imaging devices.
Another object of the present invention is to provide an overcoat process allowing the widest and most forgiving process windows for color filters and semiconductor integration reproducibility, high reliability, and, consequently maximum process and package yield.
A further object of the present invention is to obviate topographical step variations, non-planarity and surface roughness problems encountered with conventional Prior Art formation sequences. Prior Art is well known to have step-height or steric effect variations between R/G/B layers and results in departures from designer's specifications in transmittance color-balance.
Avoidance of the specific color pixel lifting problem is a still further object of the present invention.
To practice the method of the present invention, conventional microelectronic fabrication techniques using photolithographic materials, masks and etch tools are employed: in succession the array of pn-junction photodiodes is patterned with impurity dopants diffused or ion-implanted, electrically isolated, and planarized over. In the present invention, the colors only process is disclosed wherein color filters are geometrically patterned to assemble primary green, red, and blue color filters formed by the addition of suitable dyes or pigments appropriate to the desired spectral transmissivity to be associated with specified photodetector coordinate addresses in the imager matrix and the algorithm for synthetic color image reconstruction. The microlens process steps have been eliminated in the colors only process. A final specified planarization layer is applied directly above the color filter layer to complete the colors only process. The flat top surface is optimal for the package dicing and final cleaning treatment steps, minimizing particle residues and maximizing product final yield.